Amazon, Google, Facebook, Tencent, Twitter, and Microsoft are betting big on chatbots (bots), and the tech media is abuzz. This strategy makes perfect sense as tech industry heavyweights build out their ecosystems for commerce and customer support. Along with advances in conversational technologies and artificial intelligence, bot technology is getting a big shot in the arm and becoming mainstream.
While this is great for the bot industry, my concern is that we'll see a large number of poorly designed and ineffective bots hit the market. Companies will experiment with the technology without really knowing how to incorporate the right elements needed for a successful bot deployment. A spate of really bad bot deployments over the next couple years would cause a great deal of frustration and turn many consumers off.
That's important because we've seen time and time again how unforgiving customers can be if their first experiences are poor – it takes years for them to regain the confidence in a technology. As the tech industry, we have a responsibility to ensure that we prevent consumer headaches and avoid "do-overs." As such, we urge businesses considering bots to consider the following:
CHOOSE A BOT WITH INTELLIGENCE
Not all bots are created equal. The term "bot" seems to have become an umbrella term recently. It now includes everything from very basic, dumbed down automated Q&A to the more sophisticated enterprise virtual agents that understand natural language, maintain context and scale to a large number of users. Some analysts, like @Dan Miller and @Brian Manusama are hesitant to even classify virtual agents as bots. This is because most bots (with the exception of virtual agents) are nothing more than general purpose tools that only handle basic Q&A or chime in with useless phrases such as "can I help you book a flight?"
Generic bots are simply not designed to serve businesses, as they lack elements such as customer journey pathways, hook-ins to the right knowledge management systems or the ability to understand natural language. If they can't do that, how can they possibly understand customer intent? Customizing generic bots to fit the business is where most enterprises will struggle. Businesses with high volumes of customer interaction and complex business processes require a level of sophistication that draws from a variety of data sources to make the bot effective.
INTEGRATE THE BOT WITH AGENT-ASSISTED CHAT EXPERIENCES
Integrating a bot with chat is not hard. It can be as simple as putting a chat button in the bot interface. However, doing it effectively for the optimal customer experience requires accessing and embedding intelligence. A bot should be designed to easily troubleshoot the most common problems and escalate anything more complex to a human agent. It is in this area that investments have been short sighted. Almost all bot deployments to date, and those that will surface in the market in coming years, will not be intuitive as the customer escalates from a bot to a chat experience with a live agent. The context will not be there and the experience will be disjointed.
The seamless handoff from a bot to an agent, with context, is key to creating a successful customer engagement strategy. An intelligent bot that understands the customer's journey, intent, and behavior (and react accordingly) is just one piece of the puzzle. The other is being able to evolve quickly. This comes from being able to analyze chat transcripts between customer and agent and use the findings to improve the bot interaction through an infusion of knowledge in each interaction. Towards this end, the bot and agent need to draw from the same well of information.
FUTURE-PROOF YOUR INVESTMENTS IN BOTS
The reality is that channels come and go. How soon will it be before enterprises look to support customer engagement through augmented and virtual reality? Just as we've seen with other types of technology, the popularity of channels will wax and wane, so bots must be designed to stand the test of time and transcend channels. The notion of a universal concierge to front-end all digital engagement is here to stay, so you should build towards it.
In order to do this, don't just stand up a bot in one channel, but rather create a long term strategy that mitigates risk by building on a single, reusable service-oriented platform. Do it in such a way that it straddles engagement channels and that intelligence and interfaces are abstracted and invoked as a service. Towards this end the platform should be able to abstract:
- Natural language (text and voice)
- Interfaces (graphical, voice, multimodal)
- Data (interaction, unstructured and enterprise systems of record)
- Prediction (ability to identify customer intent)
These elements are constantly evolving, and incorporating them distinguishes bots that can actually drive business outcomes and improve engagement. By segregating the presentation and intelligence layers, businesses can sustain a competitive advantage in time to market, resources, and costs. This same single, reusable platform can be used to power other customer touchpoints like voice, mobile and chat, as well as others that will emerge in the future.
SUCCESSFUL BUSINESSES DO ALL-OF-THE-ABOVE
Some companies are doing it right, and we're working with leading companies in all industries to roll out intelligent chatbot technology. Here are just a few examples:
- Retail: A large U.S. retail chain is preparing to roll out as system to let consumers receive order notifications, receipts and shipping updates, all through a single messaging platform.
- Travel: A major airline is preparing to roll out a system for handling the entire flight booking life cycle including booking confirmations, reminders when check-in opens, getting a boarding pass and receiving flight status updates.
- Entertainment: An online entertainment service will soon offer service updates, outage notifications and alerts to update credit card info before it expires.
In all of these cases, consumers will be able to ask questions that can be handled by a virtual agent, or escalated to a human if needed.
Remember – A customer does not want to engage with a bot simply for the sake of using cool technology. The novelty of a bot providing simple, but unhelpful answers will fade quickly. What consumers want is technology that works for them: help pay a bill, return an item, book a hotel. The businesses that get it will avoid false starts and do it right the first time to make their bot one that engages customers in the right way, with immediacy, context and the intelligence to know what to do next. If done right, bots can be a lasting future pillar of your organization's customer engagement strategy.